Testing tick data for normality

[GKonheiser]

hi, does anyone know of a method for testing tick data for normality?

[NinjaTrader_DaveI]

Hello,

Can you please clarify exactly what you are looking to do? Are you looking to test tick data for outliers, then smooth out the data?

[GKonheiser]

Hi,

I want to test for a normal distribution around the mean for tick data. I can do it with R but I want to do it on the fly while a strategy is running.

[NinjaTrader_DaveI]

Yes, you can do this. You will probably want to add a secondary data series with a 1-tick interval, then you can perform any mean and standard deviation calculations that you wish on any portion of the data set. For more information on adding and accessing a secondary data series, please see the link below:

http://www.ninjatrader.com/support/h....html?add3.htm

There are a few useful items to note in the script sample listed in that page, including Closes and BarsInProgress. You can find more information about these in the links below:

http://www.ninjatrader.com/support/h...tml?closes.htm

http://www.ninjatrader.com/support/h...inprogress.htm

[koganam]

Hi,

I want to test for a normal distribution around the mean for tick data. I can do it with R but I want to do it on the fly while a strategy is running.

I can tell you with near certainly that price data will not have a Normal/Gaussian distribution. It has a distribution closer to a sine wave distribution.

Intuitively that makes sense, as we have a wave phenomenon, which Fourier Analysis tells us can be represented by a weighted sum of an infinite series of sines and cosines. In actuality John Ehlers has done some actual measurements and confirms this. more or less. Check out the details on the MESA software site.